Details
- Google AI announced Gemini 3.5 Live Translate, a new audio model focused on live speech-to-speech translation.
- The system supports over 70 languages, enabling real-time communication across a wide range of global users.
- Translations begin as soon as the speaker starts talking, with the model streaming output continuously rather than waiting for long pauses.
- The model is designed to keep translating while listening to what the user says next, reducing delays and conversational friction.
- Google emphasizes that the experience minimizes awkward pauses and choppy turn-taking, aiming for more natural conversational flow.
- Gemini 3.5 Live Translate builds on Google's broader work in AI-powered translation across Google Translate and Gemini models.
- A linked blog post provides technical and product details on how the model works and where it is initially available.
- The announcement signals deeper integration of advanced Gemini models into everyday language tools and real-time communication scenarios.
Impact
By introducing Gemini 3.5 Live Translate, Google strengthens its position in real-time multilingual communication and narrows the gap between human conversation and machine-mediated translation. Streaming, low-latency speech-to-speech translation across 70+ languages could make Google’s ecosystem more attractive for international callers, travelers, and businesses that rely on cross-language voice interactions.
